Abstract
Cancer remains a major health issue in NZ and worldwide. Conventional treatments such as surgery, radiation and chemotherapy have limited ability to treat cancers and cause unwanted adverse effects. In the past few decades, immunotherapies such as immune checkpoint inhibitor therapy and adoptive cell transfer have shown promising clinical evidence in treating cancers. Chimeric antigen receptor (CAR) T cell therapy has achieved remarkable success in treating relapse of refractory hematological cancers. However, current CAR T cell treatments are technically and financially challenging and are associated with potentially life-threatening adverse effects such as cytokine release syndrome, neurotoxicity and on-target-off-tumor toxicity.
In this thesis universal CAR (UniCAR) T cell therapy was investigated as a promising approach in overcoming the challenges encountered while translating CAR T cell therapy into solid tumors. The central hypothesis was that UniCAR therapy would improve the safety profile and broad applicability of CAR T cell-based treatment in solid tumors. UniCAR recognize a generic switch or tag (antibody-based or small molecule-based), in which a universal tag is linked to specific tumor-targeting ligand(s). The advantage of this approach is that only one UniCAR DNA construct needs to be developed and this can be used to transduce T cells from any patient. These cells can then be used in combination (simultaneously or sequentially) with many different specific tumor-targeting ligand conjugated tag constructs, to treat cancer expressing different tumor-specific or -associated antigens. An additional innovation investigated in this thesis to reduce on-target-off-tumor toxicity was the development of stimuli-responsive masked tag constructs (pro-tags, analogous to prodrugs). These masked constructs are not able to bind to tumor-associated antigens on healthy tissues but are unmasked in the tumor microenvironment (TME) to allow tumor specific binding. The emphasis of this thesis was on the design and synthesis of small-molecule based tags, which included glucose-biotin and folate-biotin masked and unmasked constructs. The biotin moiety was included to act as a ligand for the universal streptavidin (SA) CAR and was attached to glucose or folate, the targeting ligands for GLUT1 and folate receptor-α (FRα), that are overexpressed by cancer cells. Small molecule-based tags were hypothesized to be a better option in directing UniCAR therapy due to better penetration into the TME and the nature of the synapse formed between the CAR T cell and the cancer cell. The design of these pro-tags was such that the stimuli-responsive triggerable group was unmasked upon exposure to reductive endogenous hydrogen sulfide (H2S) and oxidative hydrogen peroxide (H2O2), stimuli upregulated in the TME.
The synthesis of glucose-biotin tags was developed and optimized to allow the conjugation of biotin to glucose at C3 or C4, which would not affect the binding of the glucose-biotin conjugate to GLUT1. A series of synthetic steps that involved protection and deprotection of hydroxyl functionalities allowed the attachment of a stimuli-responsive triggerable group at C1, which has been reported to block the binding of glucose to GLUT1, even with the installment of a small substituent. An elongated self-immolative linker containing an electron-cascade elimination linker and cyclization spacer was tethered to triggerable functional group at the C1 position. This was expected to undergo spontaneous 1,6-elimination electron cascade and cyclization cleavage after the conversion of the stimuli-responsive mask into its corresponding electron-donating derivative. Two stimuli-responsive glucose-biotin pro-tags, a nitroaromatic (2.3) and an arylboronate moiety (2.4), were developed and synthesized. These would be responsive to reductive (e.g. hydrogen sulfide, H2S) and oxidative (e.g. hydrogen peroxide, H2O2) stimuli respectively, allowing the self-immolative unmasking of glucose in the glucose-biotin pro-tags. The activation of stimuli-responsive triggerable functional groups by chemical (Zn/AcOH and H2S) and enzymatic reduction were assessed by HPLC-monitored activation studies. The nitroaromatic group of 2.3 could be activated using Zn/AcOH, with the nitroaromatic group converted to its corresponding amine (aniline) group. However, the cyclization spacer did not cyclize, presumably due to the protonation of the nitrogen atom on the primary amine of the spacer, which impacted the reactivity of this amine group to undergo self-immolation. The nitro-functionalized 2.3 was fully activated by H2S (with NaSH as a source) and led to subsequent self-immolative release of glucose-biotin tag 2.1. Enzymatic reduction with nitroreductase (NTR) failed to catalyze the bioconversion of nitroaromatic group to its aniline group likely due to 2.3 being a poor fit for the active site of NTR. The arylboronate-functionalized 2.4 was found to rapidly converted to its phenolic derivative upon exposure to 9.8 mM H2O2, and underwent a two-step self-immolation reaction sequence to give the unmasked glucose-biotin tag 2.1.
A folate-biotin tag was developed due to the high affinity of folate for folate receptor-α (FRα) and overexpression of FRα on the cancer cells. Unmasked folate-biotin tags were synthesized using two coupling methods; a standard amide coupling and click-chemistry. It was hypothesized that the attachment of a bulky substituent to the folate skeleton at N10 would prevent the effective binding of folate to the FRα. The direct attachment of a masking group to folate was unsuccessful, which was proposed to be due to steric hindrance that limited the substitution reaction. The synthesis of masked folate-biotin pro-tags was achieved through de novo synthesis, with a key intermediate produced by attaching a stimuli-responsive group to N10 through a carbamate functionality, which is stable in blood but can be driven thermodynamically to unmask upon activation by a specific stimulus. A non-triggerable masked folate-biotin pro-tag 3.34 to use as a control was also synthesized, demonstrating the possibility of attaching a bulky substituent at N10. Although a key intermediate with an azide functionalized triggerable group attached to folate was successfully synthesized, the reaction failed at the penultimate step. This was due to difficulties in purifying sufficient amounts of reaction intermediates, highlighting a potential limitation of this synthetic route.
In vitro biological evaluation of the biotin tags and pro-tags (i.e. glucose-biotin constructs and folate-biotin constructs) was performed and all constructs were found to be non-toxic. The presence of targeted antigens on EGFR.F10.EGFR melanoma cells was confirmed. A killing assay was then established utilizing two universal CARS, an mSA2 CAR that would recognize biotin and an anti-FITC CAR that was used as a control. The Platinum-E packaging cell line was transfected to produce mSA2 or anti-FITC retroviruses, which were then used to transduce actively dividing murine T cells to produce desired mSA2 and anti-FITC CAR T cells. Both the glucose-biotin tag 2.1, and anti-EGFR-biotin (positive control) were able to direct killing of the universal mSA2 CAR T cells. Killing was not detected in the absence of glucose-biotin tags or when glucose-biotin tags were used in combination with anti-FITC CAR T cells, further supporting the specificity of universal CAR T cell killing.
To summarize, the the overall goal of this thesis was to improve the safety profile and increase the applicability of CAR T cell therapy by developing a universal CAR T therapy that utilized an inactive tumor ligand-tag prodrug for tumor targeting. As an initial proof-of-concept, it was found that glucose-biotin constructs were able to direct the specific killing of mSA2 CAR T cells. Successful activation of biotin pro-tags conjugated to glucose (i.e., nitro-functionalized 2.3 and arylboronate-functionalized 2.4) demonstrated the potential of using these pro-tags to improve the safety profile of this therapy, through selective unmasking in the TME and thereby redirecting universal CAR T cells to exhibit cytotoxicity towards cancer cells, but not healthy cells.